WO2019113776A1

WO2019113776A1 - Face and voiceprint-based payment authentication method, and terminal

Info

Publication number: WO2019113776A1
Application number: PCT/CN2017/115617
Authority: WO
Inventors: 张炽成; 唐超旬
Original assignee: 福建联迪商用设备有限公司
Priority date: 2017-12-12
Filing date: 2017-12-12
Publication date: 2019-06-20
Also published as: CN108124488A

Abstract

Provided are a face and voiceprint-based payment authentication method, and a terminal. The method comprises the following steps: determining, according to facial information, whether facial image information of facial information contains high-frequency information greater than a preset first threshold; determining, according to voiceprint information, whether the voiceprint information contains a high-frequency component greater than a preset second threshold; and if both determination results are negative, performing payment authentication according to the facial image information and the voiceprint information. In the present invention, whether a user is in a compelled state can be effectively determined according to facial image information and voiceprint information, thereby ensuring a more secure and reliable payment process. The method combines face and voiceprint recognition techniques, thereby greatly reducing risks of unauthorized payments, and enhancing transaction security and reliability.

Description

Payment authentication method and terminal based on face and voiceprint

Technical field

The present invention relates to the field of electronic payment technologies, and in particular, to a payment authentication method and a terminal based on a face and a voiceprint.

Background technique

With the continuous development of Internet technology, online shopping through smart mobile terminals has become an indispensable part of people's lives, which is also very convenient for people's lives. Since online shopping involves sensitive information of users, it requires a more secure payment authentication method when shopping online and making payment. The current payment authentication method mainly performs payment authentication through fingerprint or face recognition, which has the following disadvantages: biometrics Easy to be stolen: Fingerprint information is easier to take when the trader touches the item, and the non-living information, the facial image information is originally public, and it is easy to steal through video or photographing; the stolen biometrics are easy to use. In the attack, the stolen fingerprint and facial information can be used to attack the payment device by making fingerprint and image synthesis techniques respectively, thereby achieving the purpose of stealing.

technical problem

The technical problem to be solved by the present invention is that the present invention provides a face and voiceprint payment authentication method and a terminal, which improves the security of payment authentication.

Technical solution

In order to solve the above technical problem, the present invention provides a payment authentication method based on a face and a voiceprint, comprising the following steps:

S1: determining, according to the face information, whether there is high frequency information greater than a preset first threshold in the face image information included in the face information; and determining, according to the voiceprint information, whether the voiceprint information is greater than a preset number a high frequency component of the second threshold;

S2: If no, the payment authentication is performed according to the face image information and the voiceprint information.

The present invention also provides a face and voiceprint based payment authentication terminal comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, the processor implementing the program to implement the following step:

Beneficial effect

The beneficial effects of the invention are:

The invention provides a payment authentication method and a terminal based on a face and a voiceprint. By determining whether there is high frequency information greater than a preset threshold in the face image information, it is possible to prevent the image information synthesized by the computer from attacking the authentication attack. Problem (Because of the image synthesized by the computer, a large number of spatial hopping occurs at the edge of the face, the edge of the eye, the edge of the mouth, etc., and there is a large amount of high-frequency information corresponding to the frequency domain); at the same time, the voiceprint is judged. Whether there is a high-frequency component greater than a preset second threshold in the information, which can prevent payment authentication by the spliced recording information (the spliced recording information has a high-frequency component in the splicing section), and the above method is implemented in the face and High-frequency detection is added to the voiceprint recognition to prevent face or recording synthesis attacks, and face recognition is combined with relatively difficult to disguise voiceprints for payment authentication, which can effectively avoid camouflage attacks and make payment more secure.

DRAWINGS

1 is a schematic diagram showing main steps of a face and voiceprint based payment authentication method according to an embodiment of the present invention;

2 is a schematic structural diagram of a face authentication and voiceprint based payment authentication terminal according to an embodiment of the present invention;

Label description:

1, memory; 2, processor.

Embodiments of the invention

Referring to FIG. 1, the present invention provides a payment authentication method based on face and voiceprint, which includes the following steps:

As can be seen from the above description, the present invention provides a payment authentication method based on a face and a voiceprint. By determining whether there is high frequency information greater than a preset threshold in the face image information, it is possible to prevent image information pairing by computer synthesis. The problem of authentication attack (due to the image synthesized by the computer, a large number of spatial hopping occurs at the edge of the face, the edge of the eye, the edge of the mouth, etc., and there is a large amount of high frequency information corresponding to the frequency domain); Determining whether there is a high-frequency component greater than a preset second threshold in the voiceprint information, and preventing payment authentication by the stitched recording information (the stitched recording information has a high-frequency component in the splicing section), and the method is implemented High-frequency detection is added to face and voiceprint recognition to prevent facial or recording synthesis attacks, and face recognition is combined with relatively difficult to disguise voiceprints for payment authentication, which can effectively avoid camouflage attacks and make payment more secure.

Further, before the S1, the method further includes:

S01: displaying face specified action information required for payment verification and designated text information requiring voice input;

S02: Simultaneously collecting voice information while simultaneously generating voice information; the face information includes face video information and face image information;

S03: determining whether the face motion in the face video information is consistent with the specified action, and determining whether the text corresponding to the voiceprint information is consistent with the text corresponding to the specified text information;

S04: If they are all consistent, step S1 is performed, otherwise the payment authentication fails.

It can be seen from the above description that the collected face information is the face video information and the face image information collected by the camera device when the user performs the corresponding face action according to the specified action information displayed; the collected voiceprint information is displayed by the user according to the display. The specified text information is used to input the voiceprint information obtained by voice input, and the user does not know the specified action information and the specified text information in advance, and the security of the payment verification is improved by the above verification method.

Further, the S03 and the S04 further include:

Determining whether the audio information in the lip language information and the voiceprint information in the face video information is synchronized, and if not, the payment authentication fails;

If the synchronization is performed, it is determined whether the first text information corresponding to the lip language information is consistent with the specified text information.

It can be seen from the above description that, by the above method, it is possible to ensure that the lip language information is synchronized with the audio information, and that the lip language information corresponds to the specified text information, thereby making the payment verification more secure and reliable.

Further, the S02 and the S03 further include:

The face information and the voiceprint information are separately subjected to noise reduction and filtering processing.

As can be seen from the above description, the accuracy of data processing can be improved by the above method.

Further, the face specified action information required for the payment verification and the specified text information requiring voice input are displayed in the S01:

The face specified action information and the specified text information requiring voice input are randomly generated, and the specified action information and the specified text information are displayed.

As can be seen from the above description, by the above method, it is possible to prevent the fraudulent use of the face information at the time of last support verification for payment verification, and at the same time, it is possible to prevent the criminals from recording the voice input when the user pays, thereby ensuring the security of the payment.

Further, if the face information or the voiceprint information fails to be collected within the preset time, the new specified action information and the new designated text information are randomly displayed, and the face information and the voiceprint information are re-acquired.

As apparent from the above description, by the above method, it is possible to prevent the payment authentication information from being stolen, and the security of the payment is further improved.

Further, before the S2, the method further includes:

Acquiring a plurality of first face image information in a state of being stressed, and calculating a first feature parameter of each first face image information;

Fitting all the first feature parameters to obtain a first mathematical model between the stressed state and the first feature parameter;

Acquiring a plurality of first voiceprint information in a state of being stressed, and calculating a second feature parameter of each first voiceprint information;

Fitting all the second characteristic parameters to obtain a second mathematical model between the stressed state and the characteristic parameters.

It can be seen from the above description that the first mathematical model of the voiceprint information under the stress state and the second mathematical model of the face image information under the stress state can be established by the above method, so that the state of the user can be accurately determined subsequently. Increase the security of payment.

Further, before the S2, the method further includes:

Calculating a face feature parameter of the face image information in the obtained face information;

Calculating the voiceprint characteristic parameters of the obtained voiceprint information;

Determining whether the user is in a coerced state according to the first mathematical model and the face feature parameter;

According to the second mathematical model and the voiceprint feature parameter, it is determined whether the user is in a state of being coerced.

As can be seen from the above description, by the above method, it is possible to accurately determine whether or not the use is in a state of being coerced, thereby preventing the user from being robbed of user information by the criminal in the state of being coerced, causing a huge loss to the user.

Further, the S2 is specifically:

If yes, encrypting the face feature parameter and the voiceprint feature parameter, and sending the parameter to the server, so that the server performs a significant analysis on the first face feature parameter corresponding to the face feature parameter and the pre-stored face information, and The first voiceprint feature parameter corresponding to the voiceprint feature parameter and the pre-stored voiceprint information is subjected to significant analysis to obtain a significant analysis result;

Based on the result of the significance analysis, it is determined whether the payment authentication is passed.

It can be seen from the above description that the face feature parameters and the voiceprint feature parameters are encrypted and transmitted during the transmission process, which can prevent the user data from being stolen and cause a certain economic loss to the user; and at the same time, the saliency analysis can accurately determine Whether the feature parameters corresponding to the face feature parameters and the pre-stored face information match, and can accurately determine whether the feature parameters corresponding to the voiceprint feature parameters and the pre-stored voiceprint information match; the above-mentioned double verification method improves the security of payment authentication.

Further, the S1 is specifically:

Calculating image frequency domain information corresponding to the face image information, and determining whether there is high frequency information greater than a preset threshold in the image frequency domain information;

And calculating, according to the voiceprint information, a high frequency component set, and determining whether there is a high frequency component greater than a preset threshold in the high frequency component set.

It can be seen from the above description that, by the above method, it is possible to accurately calculate whether the voiceprint information includes a high frequency component, to prevent the synthesized recording from being used for payment verification, and to accurately determine whether the high frequency information is included in the face image information, The face image information synthesized by the computer is prevented from being used for payment verification, and the above method improves the security of payment.

Further, before the S1, the method further includes:

Performing an authentication judgment on the preset transaction terminal and the server, and if the authentication fails, the payment authentication fails, and the transaction is ended;

If the authentication is successful, acquiring current location information of the transaction terminal;

Encrypting the current location information to obtain location encryption information;

Sending the location encryption information to the server, so that the server saves the location encryption information in preset security log information.

As can be seen from the above description, the location information in which the transaction is stored during the transaction process makes the transaction location traceable.

Referring to FIG. 2, the present invention provides a face and voiceprint based payment authentication terminal, comprising a memory 1, a processor 2, and a computer program stored on the memory 1 and operable on the processor 2, characterized in that The processor 2 implements the following steps when executing the program:

Further, the above-mentioned face and voiceprint-based payment authentication terminal further includes:

Further, the above-mentioned face and voiceprint-based payment authentication terminal further includes: between S03 and S04:

Further, the above-mentioned face and voiceprint-based payment authentication terminal further includes: between S02 and S03:

Further, the above-mentioned face and voiceprint-based payment authentication terminal, the face specified action information required for displaying the payment verification in S01 and the specified text information requiring voice input are specifically:

Further, the payment authentication terminal based on the face and the voiceprint is configured to randomly display the new specified action information and the new designated text information if the face information or the voiceprint information fails to be collected within a preset time. And re-collect face information and voiceprint information.

Further, the above-mentioned face and voiceprint-based payment authentication terminal, before the S2, further includes:

Further, the above-mentioned face and voiceprint-based payment authentication terminal, the S2 is specifically:

Referring to FIG. 1, a first embodiment of the present invention is:

The invention provides a payment authentication method based on face and voiceprint, comprising the following steps:

S0: randomly generating face specified action information and designated text information requiring voice input, displaying the specified action information and designated text information; collecting face information while simultaneously generating voiceprint information; the face information including face video Information and face image information; after performing noise reduction and filtering processing on the face information and the voiceprint information respectively, determining whether the face motion in the face video information is consistent with the specified action, and determining the text corresponding to the voiceprint information Whether the characters corresponding to the specified text information are consistent; determining whether the lip information in the face video information and the audio information in the voiceprint information are synchronized, if not, the payment authentication fails; if the synchronization, the lip information is determined to be corresponding Whether the first text information is consistent with the specified text information; if they are consistent, step S1 is performed, otherwise the payment authentication fails;

Wherein, if the face information and the voiceprint information collection fails within the preset time, step S0 is re-executed;

Embodiment 2 of the present invention is:

Acquiring a plurality of first face image information in a state of being stressed, calculating a first feature parameter of each first face image information; fitting all the first feature parameters to obtaining a stressed state and a first feature parameter The first mathematical model between;

Wherein "fitting all the first characteristic parameters to obtain a first mathematical model between the stressed state and the first characteristic parameter" is specifically:

Reading the first feature parameter corresponding to each sample (face image information), fitting all the first feature parameters by deep learning convolutional neural network, ie defining neural network, collecting original data, classifying training And correcting, outputting the result, obtaining a first mathematical model between the stressed state and the first characteristic parameter;

Acquiring a plurality of first voiceprint information in a state of being stressed, calculating a second feature parameter of each first voiceprint information; fitting all second feature parameters to obtain a relationship between the stress state and the feature parameter Second mathematical model;

Wherein "fitting all the second characteristic parameters to obtain a second mathematical model between the stressed state and the first characteristic parameter" is specifically:

Reading the second characteristic parameter corresponding to each sample (soundprint information), fitting all the second characteristic parameters by deep learning convolutional neural network, ie defining neural network, collecting original data, classifying training, Correcting and outputting a result, obtaining a second mathematical model between the stressed state and the second characteristic parameter;

In the payment, the default transaction terminal and the server are first authenticated. If the authentication fails, the payment authentication fails and the transaction is terminated.

If the authentication succeeds, acquiring current location information of the transaction terminal; encrypting the current location information to obtain location encryption information; and transmitting the location encryption information to a server, so that the server saves the location encryption information in a preset Security log information;

After the authentication succeeds, the face specified action information and the specified text information requiring voice input are randomly generated, and the specified action information and the specified text information are displayed; while the face information is collected, the voiceprint information is simultaneously; the face information is The method includes the face video information and the face image information; after performing noise reduction and filtering processing on the face information and the voiceprint information, determining whether the face motion in the face video information is consistent with the specified action, and determining the voiceprint Whether the text corresponding to the information is consistent with the text corresponding to the specified text information; determining whether the audio information in the lip information and the voiceprint information in the face video information is synchronized, if not synchronized, the payment authentication fails; if it is synchronized, it is determined Whether the first text information corresponding to the lip language information is consistent with the specified text information; if there is an inconsistency, the payment authentication fails, otherwise the following steps are performed:

Determining, according to the face information, whether there is high frequency information greater than a preset first threshold in the face image information included in the face information; and determining, according to the voiceprint information, whether the voiceprint information is greater than a preset second threshold High frequency component;

Calculating a face feature parameter of the face image information in the obtained face information; calculating a voiceprint feature parameter of the obtained voiceprint information; determining, according to the first mathematical model and the face feature parameter, whether the user is under stress State; determining whether the user is in a coerced state according to the second mathematical model and the voiceprint feature parameter.

Referring to FIG. 2, the third embodiment of the present invention is:

The present invention provides a face and voiceprint based payment authentication terminal comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the processor executes the program The following steps are implemented:

Embodiment 4 of the present invention is:

Embodiment 5 of the present invention is:

The invention provides a POS machine, comprising an MCU (micro control module), a camera, a microphone and a liquid crystal screen, wherein the MCU is electrically connected to the camera, the microphone and the liquid crystal screen respectively;

1) Before leaving the factory, the POS software undergoes extensive machine learning training, including tens of thousands of face information under normal emotions and coerced face information, as well as tens of thousands of normal voice messages and voice information under stress. The recognition software fits into a calculation formula by reading the specific parameters of each training sample and fitting it into a calculation formula by using the deep learning convolutional neural network for all training samples (the main steps are: defining the neural network, collecting the original data, and classifying Training, correction, and output results), the relationship between these parameters and emotions is derived, so that the software for face recognition and voiceprint recognition has the ability to identify whether the source of information is under coercion.

2) POS and transaction background authentication: If the authentication fails, it means that the POS has no transaction authority and ends the transaction; if the authentication is successful, it means that the POS has the transaction authority, and the encryption of the POS and the transaction background is enabled. The wireless module encrypts and uploads the current base station location at this time, and saves it as the content of the security log in the transaction background.

3) The internal MCU of the POS machine randomly generates the written text information, and prompts the trader to use the microphone to read the corresponding text information through the LCD screen, and the MCU randomly generates the face specified action (including blinking, opening mouth, turning head, etc.) through the LCD screen. The user is prompted to collect the face information of the specified action through the camera.

4) The camera collects the face information of the trader, and at the same time, the microphone collects the voice information of the trader. While collecting sound information, face information is still collected for lip language calculation.

5) The MCU preprocesses the face information, including noise reduction and normalization processing.

6) The MCU checks the legality of the face information, including checking whether the face action is consistent with the prompt, checking whether there is high frequency information exceeding the threshold, whether it is not in a nervous fear, whether the lip language is consistent with the prompt information, and whether the lip language is Synchronized with the recorded information, if the check fails, it is judged as illegal information, rejected, and the transaction is terminated.

7) The MCU calculates the feature values of the face information, including the geometric features of the eyes, nose, mouth, and the like of the face.

8) The MCU encrypts the face information feature value and transmits it to the transaction background.

9) The transaction background analyzes the feature value of the uploaded face information and the face information of the cardholder reserved by the bank, and judges whether the transaction is allowed according to the analysis result: if the significance is insufficient, the face recognition fails, and the POS is notified. End the transaction; if the significance is obvious, it indicates that the face recognition is successful, and the POS is allowed to allow the transaction.

10) The MCU preprocesses the sound information, including noise reduction and the like.

11) The MCU checks the legality of the sound, including checking whether the sound content is consistent with the prompt, checking whether there is high frequency information exceeding the threshold, whether the emotion is not in a nervous or fearful mood, and if the check fails, it is judged as illegal information and rejected. End the transaction.

12) The MCU calculates the voiceprint feature information, that is, MFCC (Mel Frequency Cepstrum Coefficients).

13) The MCU encrypts the voiceprint feature value and transmits it to the transaction background.

14) The transaction background analyzes the uploaded voiceprint feature value and the cardholder's voiceprint information reserved by the bank: if the significance is insufficient, it indicates that the voiceprint recognition fails, and the POS is notified to end the transaction; if the significance is obvious, then Indicates that voiceprint recognition is successful, performs background transactions, and informs the POS of the transaction result.

15) The POS machine will prompt the trader to the transaction result.

In summary, the present invention provides a method and a terminal for payment authentication based on a face and a voiceprint. By determining whether there is a high frequency component greater than a preset second threshold in the voiceprint information, the recorded information can be prevented from being stitched. Perform payment authentication (spliced recording information has high-frequency components in the splicing segment); in the image synthesized by computer, a large number of spatial hopping occurs at the edge of the face, the edge of the eye, the edge of the mouth, etc. In the frequency domain, a large amount of high-frequency information exists. Therefore, by determining whether there is high-frequency information larger than a preset threshold in the face image information, it is possible to prevent the problem of the authentication authentication attack by the computer-synthesized image information; Face information and voiceprint information can effectively determine whether the user is in a state of coercion, making payment more secure and reliable. The combination of face and voiceprint can greatly reduce the risk of misappropriation and enhance transaction security and reliability. The cardholder's face information and voiceprint information are transmitted to the server as encrypted information through encryption. Only the one-way uplink transmission of the characteristic parameters is allowed to avoid the leakage of sensitive information. The above method realizes adding high frequency detection in face and voiceprint recognition to prevent face or recording synthesis attack, and performs face authentication combined with relatively difficult to disguise voiceprint for payment authentication, which can effectively avoid camouflage attack and make payment safer.

The above is only the embodiment of the present invention, and thus does not limit the scope of the patent of the present invention. All equivalent transformations made by using the specification and the drawings of the present invention, or directly or indirectly applied to other related technical fields, include the same. Within the scope of the patent protection of the present invention.

Claims

A payment authentication method based on face and voiceprint, characterized in that it comprises the following steps:

S1: determining, according to the face information, whether there is high frequency information greater than a preset first threshold in the face image information included in the face information; and determining, according to the voiceprint information, whether the voiceprint information is greater than a preset number a high frequency component of the second threshold;

S2: If no, the payment authentication is performed according to the face image information and the voiceprint information.
The face and voiceprint-based payment authentication method according to claim 1, wherein before S1, the method further comprises:

S01: displaying face specified action information required for payment verification and designated text information requiring voice input;

S02: Simultaneously collecting voice information while simultaneously generating voice information; the face information includes face video information and face image information;

S03: determining whether the face motion in the face video information is consistent with the specified action, and determining whether the text corresponding to the voiceprint information is consistent with the text corresponding to the specified text information;

S04: If they are all consistent, step S1 is performed, otherwise the payment authentication fails.
The face and voiceprint-based payment authentication method according to claim 2, wherein the S03 and the S04 further comprise:

Determining whether the audio information in the lip language information and the voiceprint information in the face video information is synchronized, and if not, the payment authentication fails;

If the synchronization is performed, it is determined whether the first text information corresponding to the lip language information is consistent with the specified text information.
The face and voiceprint-based payment authentication method according to claim 2, wherein the S02 and the S03 further comprise:

The face information and the voiceprint information are separately subjected to noise reduction and filtering processing.
The face and voiceprint-based payment authentication method according to claim 2, wherein the face specified action information required for displaying the payment verification in S01 and the specified text information requiring voice input are specifically :

The face specified action information and the specified text information requiring voice input are randomly generated, and the specified action information and the specified text information are displayed.
The face and voiceprint-based payment authentication method according to claim 5, wherein if the face information or the voiceprint information fails to be collected within a preset time, the new specified action information and the new one are randomly displayed. Specify text information and re-collect face information and voiceprint information.
The face and voiceprint-based payment authentication method according to claim 1, wherein before S1, the method further comprises:

Acquiring a plurality of first face image information in a state of being stressed, and calculating a first feature parameter of each first face image information;

Fitting all the first feature parameters to obtain a first mathematical model between the stressed state and the first feature parameter;

Acquiring a plurality of first voiceprint information in a state of being stressed, and calculating a second feature parameter of each first voiceprint information;

Fitting all the second characteristic parameters to obtain a second mathematical model between the stressed state and the characteristic parameters.
The face and voiceprint-based payment authentication method according to claim 7, wherein the S2 before:

Calculating a face feature parameter of the face image information in the obtained face information;

Calculating the voiceprint characteristic parameters of the obtained voiceprint information;

Determining whether the user is in a coerced state according to the first mathematical model and the face feature parameter;

According to the second mathematical model and the voiceprint feature parameter, it is determined whether the user is in a state of being coerced.
The face and voiceprint based payment authentication method according to claim 8, wherein the S2 is specifically:

If yes, encrypting the face feature parameter and the voiceprint feature parameter, and sending the parameter to the server, so that the server performs a significant analysis on the first face feature parameter corresponding to the face feature parameter and the pre-stored face information, and The first voiceprint feature parameter corresponding to the voiceprint feature parameter and the pre-stored voiceprint information is subjected to significant analysis to obtain a significant analysis result;

Based on the result of the significance analysis, it is determined whether the payment authentication is passed.
The face and voiceprint-based payment authentication method according to claim 1, wherein before S1, the method further comprises:

Performing an authentication judgment on the preset transaction terminal and the server, and if the authentication fails, the payment authentication fails, and the transaction is ended;

If the authentication is successful, acquiring current location information of the transaction terminal;

Encrypting the current location information to obtain location encryption information;

Sending the location encryption information to the server, so that the server saves the location encryption information in preset security log information.
A face and voiceprint based payment authentication terminal includes a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the processor implements the following steps when executing the program :

S1: determining, according to the face information, whether there is high frequency information greater than a preset first threshold in the face image information included in the face information; and determining, according to the voiceprint information, whether the voiceprint information is greater than a preset number a high frequency component of the second threshold;

S2: If no, the payment authentication is performed according to the face image information and the voiceprint information.
The face authentication and voiceprint-based payment authentication terminal according to claim 11, wherein the S1 before:

S01: displaying face specified action information required for payment verification and designated text information requiring voice input;

S02: Simultaneously collecting voice information while simultaneously generating voice information; the face information includes face video information and face image information;

S03: determining whether the face motion in the face video information is consistent with the specified action, and determining whether the text corresponding to the voiceprint information is consistent with the text corresponding to the specified text information;

S04: If they are all consistent, step S1 is performed, otherwise the payment authentication fails.
The face authentication and voiceprint-based payment authentication terminal according to claim 12, wherein the S03 and the S04 further comprise:

Determining whether the audio information in the lip language information and the voiceprint information in the face video information is synchronized, and if not, the payment authentication fails;

If the synchronization is performed, it is determined whether the first text information corresponding to the lip language information is consistent with the specified text information.
The face authentication and voiceprint-based payment authentication terminal according to claim 12, wherein the S02 and the S03 further comprise:

The face information and the voiceprint information are separately subjected to noise reduction and filtering processing.
The face authentication and voiceprint-based payment authentication terminal according to claim 12, wherein the face specified action information required for displaying the payment verification in S01 and the specified text information requiring voice input are specifically :

The face specified action information and the specified text information requiring voice input are randomly generated, and the specified action information and the specified text information are displayed.
The face authentication and voiceprint-based payment authentication terminal according to claim 15, wherein if the face information or the voiceprint information collection fails within a preset time, the new specified action information and the new one are randomly displayed. Specify text information and re-collect face information and voiceprint information.
The face authentication and voiceprint-based payment authentication terminal according to claim 11, wherein the S1 before:

Acquiring a plurality of first face image information in a state of being stressed, and calculating a first feature parameter of each first face image information;

Fitting all the first feature parameters to obtain a first mathematical model between the stressed state and the first feature parameter;

Acquiring a plurality of first voiceprint information in a state of being stressed, and calculating a second feature parameter of each first voiceprint information;

Fitting all the second characteristic parameters to obtain a second mathematical model between the stressed state and the characteristic parameters.
The face authentication and voiceprint-based payment authentication terminal according to claim 17, wherein the S2 before:

Calculating a face feature parameter of the face image information in the obtained face information;

Calculating the voiceprint characteristic parameters of the obtained voiceprint information;

Determining whether the user is in a coerced state according to the first mathematical model and the face feature parameter;

According to the second mathematical model and the voiceprint feature parameter, it is determined whether the user is in a state of being coerced.
The face authentication and voiceprint-based payment authentication terminal according to claim 18, wherein the S2 is specifically:

If yes, encrypting the face feature parameter and the voiceprint feature parameter, and sending the parameter to the server, so that the server performs a significant analysis on the first face feature parameter corresponding to the face feature parameter and the pre-stored face information, and The first voiceprint feature parameter corresponding to the voiceprint feature parameter and the pre-stored voiceprint information is subjected to significant analysis to obtain a significant analysis result;

Based on the result of the significance analysis, it is determined whether the payment authentication is passed.
The face authentication and voiceprint-based payment authentication terminal according to claim 11, wherein the S1 before:

Performing an authentication judgment on the preset transaction terminal and the server, and if the authentication fails, the payment authentication fails, and the transaction is ended;

If the authentication is successful, acquiring current location information of the transaction terminal;

Encrypting the current location information to obtain location encryption information;

Sending the location encryption information to the server, so that the server saves the location encryption information in preset security log information.